Summary: |
With the increasing sophistication of electricity markets, the need for real-time short - and medium-term predictions of wind power has risen . The use of tools based on linearized models or purely statistical analyses is questionable when the complexity of local orography induces strong non-linearities in the flow patterns that these techniques cannot predict. To overcome these shortcomings, fully non-linear computational fluid dynamics (CFD) codes must be used. CFD codes have one significant downside: they are computationally expensive and require long CPU times. It is the reduction of these times that this project seeks.
The objective of this project is the acceleration of two CFD codes developed in-house. The first is VENTOS, which uses a Reynolds Average Navier-Stokes (RANS) formulation. The second code is a Large Eddy Simulation (LES) to model atmospheric flows also over complex terrains. Acceleration will be sought via parallelization, advanced linear solvers and other solution strategies. |